In a graphics rendering pipeline, such as may be used to generate a two-dimensional (2D) raster representation of a three-dimensional (3D) scene, one or more pixel shaders may be implemented in order to assign colors and/or apply effects to pixels within the 2D raster representation. Especially with respect to high-resolution content, pixel shading may constitute a particularly significant consumer of graphics processing resources and/or power. One approach to reducing such processing and/or power costs is the implementation of multi-phase pixel shading. Multi-phase pixel shading may generally involve implementing different respective pixel shading rates for different sets of pixel shader inputs. Each possible pixel shading rate may embody a different underlying combination of comparative performance and quality. For example, for any particular set of pixel shader inputs, the use of a less computationally intensive coarse pixel shading rate may yield an increase in performance at a cost of some amount of image quality relative to the use of a standard pixel shading rate.
With respect to any given pixel shader, the implementation of multi-phase pixel shading will generally be most efficient when performance benefits are obtained at the least possible cost in terms quality, and/or when quality improvements are obtained at the least possible cost in terms of performance. A desirable partitioning of a set of pixel shader inputs for multi-phase pixel shading will partition the pixel shader inputs accordingly. However, manually determining a desirable partitioning for a set of pixel shader inputs can be a challenging task for developers, and may involve extensive analysis of shader usage and/or numerous trial-and-error iterations.